NegRisk Arbitrage Has Limits: New Research Changes the Math
A new research paper draws a hard line between theoretical and executable arbitrage in Polymarket's negative-risk markets, meaning some price gaps you're seeing right now may be structurally untradeable. Meanwhile, 6 new repos hit the radar today including a multi-strategy bot with embedded LLM advisory and a statistical weather arbitrage system worth serious attention.
NegRisk Arbitrage Isn't What You Think It Is
The most important finding today has nothing to do with a new bot. A research paper circulating in the alpha feed reframes how operators should think about negative-risk market pricing on Polymarket. The paper distinguishes between *payoff-space* no-arbitrage — which always holds mathematically — and *protocol-executable* no-arbitrage, which does not. The practical implication: price inconsistencies you can see in multi-outcome markets may be real but untradeable due to position transformation limitations baked into the protocol. Before you size into any NegRisk spread, you need to know which category you're in. This is required reading if you're running the `arbitrage` bot or anything touching multi-outcome markets.
---
New Bots: 6 Worth Cloning, 1 to Watch
**DRADIS** is the headline repo today. It runs a full multi-strategy stack — momentum, market-making, arbitrage, and ML — with an AI LLM advisor layer running locally via Ollama. The LLM component is advisory, not execution-critical, which is the right architecture. The breadth of strategies in a single codebase makes it a strong base for operators who want to consolidate. Clone and audit the risk sizing logic before live deployment.
**zeus** takes a narrow, well-defined approach: statistical weather arbitrage using forecast fusion and calibration across multiple weather data sources, with risk-sized execution. Weather markets on Polymarket have historically been soft targets for calibrated forecasters. This bot formalizes that edge. If you have meteorological data access, this is immediately relevant.
**arbitrage** focuses specifically on NegRisk multi-outcome markets — read the research paper above before running it. The strategy is sound in principle; the question is which specific markets expose you to the protocol-execution limitation the paper identifies.
**winning-wallet-finder** does exactly what the name says: statistical copy-trading with z-score edge calculation to identify and track high-performing wallets. No new wallets surfaced in today's scan, but the tooling is solid for ongoing surveillance.
**polymarket-codex-bot** is flagged as **monitor only**. Strategy is unclear — likely a template variant or a reskinned common bot. No confirmed malware, but the opacity warrants caution. Do not deploy without full code review.
No confirmed malware in today's batch. Six of eight new repos are worth cloning.
---
Security: Star-Farming Campaign Active
Seven security alerts today, with the dominant pattern being a coordinated Telegram copy-trading bot star-farming campaign. This has surfaced in multiple alerts, suggesting an organized push to inflate GitHub star counts on low-quality or potentially malicious bots promoted through Polymarket Telegram channels. A separate keyword-stuffed bot pack with fake stars was also flagged.
**Operational rule**: GitHub star count means nothing right now. Treat any bot surfaced through Telegram promotion as suspect until independently audited. Check commit history, contributor accounts, and whether the repo actually executes what the README claims.
---
Alpha Signals
Three research papers in the alpha feed today, no live trade signals.
Beyond the NegRisk paper, an agentic AI pipeline that clusters prediction markets by topic and identifies implicit equivalences and contradictions across contract texts is worth tracking for operators running cross-market strategies. Market fragmentation creates real hidden arbitrage — this approach automates the discovery layer.
A third paper covers LLM-based M&A deal outcome prediction from long-context technical documents. Niche, but directly applicable to anyone trading merger resolution markets.
---
Strategy Intelligence
The HN highlight about a bot that *always buys No on non-sports markets* scoring 471 points is a signal in itself. The "Nothing Ever Happens" framing resonates because it performs. Low-activity markets with inflated Yes pricing remain systematically exploitable. Simple base-rate strategies are outperforming in the current environment.
---
What to Watch
Spain's block of Polymarket and Kalshi over gambling licensing is a regulatory escalation worth monitoring for geographic spread. The Iran missile story threatening gamblers is a reminder that high-stakes political markets carry off-platform risk. Watch for any protocol changes Polymarket makes to NegRisk market mechanics in response to the arbitrage research going public.